Automated FAI Angle Measurement from ZTE MRI Matches Expert Manual Readings

A recent study available on arXiv (2608.07368) reveals that deep learning combined with geometric modeling can autonomously derive femoroacetabular impingement (FAI) angles from zero echo time (ZTE) MRI, aligning closely with expert manual assessments. While computed tomography (CT) is the standard for 3D osseous morphometry in FAI, it entails exposure to ionizing radiation and requires manual measurement. ZTE MRI effectively visualizes cortical bone and produces FAI angles consistent with CT results, but automated extraction has been previously limited. This cross-sectional study (evidence level 3) analyzed pelvic ZTE MRI from 73 individuals (average age 36.8 ± 18.5 years; 51 women, 22 men), resulting in 135 hips. A nnU-Net was utilized to segment the femur, pelvis, and three osseous landmarks based on 100 manually curated hips. Custom algorithms calculated various angles from the segmentations, and measurements from 35 test hips were compared to the average of two radiologist readings. The findings indicate that automated measurements are in agreement with expert evaluations, suggesting that ZTE MRI with deep learning could serve as a radiation-free, automated method for assessing FAI, potentially enhancing clinical efficiency and patient safety while decreasing dependence on CT and manual measurements.